Intermittent Aeration Suppresses Nitrite-Oxidizing ... - ACS Publications

Apr 27, 2017 - and Barth F. Smets*. Department of Environmental Engineering, Technical University of Denmark, Miljøvej Building 113, 2800 Kongens Lyn...
1 downloads 10 Views 574KB Size
Subscriber access provided by UB + Fachbibliothek Chemie | (FU-Bibliothekssystem)

Article

Suppression of nitrite-oxidizing bacteria in intermittently membrane-aerated biofilms: a model-based explanation Yunjie Ma, Carlos Domingo-Félez, Benedek Gy Plosz, and Barth F. Smets Environ. Sci. Technol., Just Accepted Manuscript • Publication Date (Web): 27 Apr 2017 Downloaded from http://pubs.acs.org on May 1, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Environmental Science & Technology is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 27

Environmental Science & Technology

1

Suppression of nitrite-oxidizing bacteria in intermittently

2

membrane-aerated biofilms: a model-based explanation

3 Yunjie Ma, Carlos Domingo-Félez, Benedek Gy. Plósz†, and Barth F. Smets*

4 5 6

Department of Environmental Engineering, Technical University of Denmark, Miljøvej Building 113, 2800

7

Kongens Lyngby, Denmark.

8 9 10 11

* Corresponding Author: E-mail: [email protected], Tel: +45 4525 1600 Fax: +45 4593 2850 †

Current address: Department of Chemical Engineering, University of Bath, Claverton Down, BA2 7AY Bath, United Kingdom.

1 ACS Paragon Plus Environment

Environmental Science & Technology

Page 2 of 27

12

Abstract

13

Autotrophic ammonium oxidation in membrane-aerated biofilm reactors (MABRs) can make treatment of

14

ammonium-rich wastewaters more energy-efficient, especially within the context of short-cut ammonium

15

removal. The challenge is to exclusively enrich ammonium-oxidizing bacteria (AOB). To achieve nitritation,

16

strategies to suppress nitrite-oxidizing bacteria (NOB) are needed, which are ideally grounded on an

17

understanding of underlying mechanisms. In this study, a counter-diffusion nitrifying biofilm reactor was

18

operated under intermittent aeration. During eight months operation, AOB dominated, while NOB were

19

suppressed. Based on dissolved oxygen (DO), ammonium, nitrite and nitrate profiles within the biofilm and

20

in the bulk, a 1-dimensional nitrifying biofilm model was developed and calibrated. The model was utilized

21

to explore the potential mechanisms of NOB suppression associated with intermittent aeration, considering

22

DO limitation, direct pH effects on enzymatic activities, and indirect pH effects on activity via substrate

23

speciation. The model predicted strong periodic shifts in the spatial gradients of DO, pH, free ammonia and

24

free nitrous acid, associated with aerated and non-aerated phases. NOB suppression during intermittent

25

aeration was mostly explained by periodic inhibition caused by free ammonia due to transient periodic pH

26

upshifts. Dissolved oxygen limitation did not govern NOB suppression. Different intermittent aeration

27

strategies were then evaluated for nitritation success in intermittently aerated MABRs: both aeration

28

intermittency and duration were effective control parameters.

29

2 ACS Paragon Plus Environment

Page 3 of 27

Environmental Science & Technology

30

Introduction

31

Short-cut ammonium (NH4 ) removal via nitrite (NO2 ) is more energy- and cost- efficient than traditional

32

NH4 removal via nitrate (NO3 ) due to reduced aeration and external electron donor requirements.1,2,3 This

33

process requires full nitritation (oxidation of all NH4 to NO2 ) and zero nitratation (oxidation of none of the

34

NO2 to NO3 ); in other words minimal activity of nitrite-oxidizing bacteria (NOB) and maximal activity of

35

ammonium-oxidizing bacteria (AOB). Similar conditions– with only partial nitritation- can also be exploited

36

to convert NH4 to a 50:50 mixture of NO2 and NH4 , which can then be coupled to anoxic NH4 oxidation

37

to attain even more resource efficient ammonium removal.4,5

38

Various conditions have been successfully tested to suppress NOB over AOB activity or wash-out NOB over

39

AOB biomass to attain nitritation in suspended growth systems. They include the operation of bioreactors at

40

limited dissolved oxygen (DO) concentrations,6 at high temperature combined with low solids retention

41

times,1 and at elevated free ammonia (FA) and/or free nitrous acid (FNA) concentrations.7 In all cases NOB

42

suppression or outcompetition versus AOB is based on differential growth kinetics. Sometimes, the proper

43

choice of system inoculum also accelerates AOB over NOB selection.8 By contrast, maintaining long-term

44

nitritation in biofilm-based reactors can be more challenging9 due to long solids retention times in biofilms

45

that interfere with outcompetition based on kinetic principles. Finding operational conditions and confirming

46

mechanisms that suppress NOB in biofilms remains a challenge. On the one hand, the existence of strong

47

spatial chemical gradients (e.g. of DO, pH and nitrogenous species) in nitrifying biofilms10 makes it difficult

48

to prescribe environmental conditions that favor AOB over NOB in the system. On the other hand, the

49

existence of multiple simultaneous chemical gradients complicates identification of the underlying

50

mechanism(s) that suppresses NOB. For example, pH and DO gradients occur simultaneously in active

51

nitrifying biofilms:11 it is difficult to unravel to what extent nitritation failure or success is associated with

52

the differential effect of oxygen (AOB and NOB having different oxygen affinities)12 or the differential

53

effects of pH (AOB and NOB responding differently to pH– as a consequence of the pH-dependent

54

maximum growth rates13,14 and the pH-dependent speciation of FA and FNA which act as both substrates and

55

inhibitors).

-

+

-

+

+

-

-

-

+

-

+

+

3 ACS Paragon Plus Environment

Environmental Science & Technology

Page 4 of 27

56

Mathematical models are one way to describe multiple processes that occur simultaneously in time and space

57

in nitrifying biofilms.15,16 A multi-species nitrifying biofilm model (MSNBM) was explicitly developed to

58

study the competition between AOB and NOB; effects of DO, pH, FA and FNA on growth kinetics were

59

incorporated in a spatially explicit way to evaluate operational conditions for NOB suppression in co-

60

diffusion biofilms.3,17 Park et al.17 showed that FA inhibition of NOB was more efficient in nascent biofilms

61

(when residual NH4 was still high), but that DO limitation was the dominant mechanism of NOB

62

suppression in established biofilms. Besides bulk DO and influent NH4 concentration, the model suggested

63

that bulk buffer capacity was another means to manipulate NOB suppression by affecting pH gradients

64

within biofilms.

65

While AOB/NOB competition in conventional co-diffusion biofilms has been studied in some detail,3,17,18

66

there are less studies on AOB/NOB competition in the context of nitritation in counter-diffusion biofilms.

67

Counter–diffusion biofilms develop in membrane-aerated biofilm reactors (MABRs), where air delivery is

68

via the biofilm base.19 MABRs have been broadly explored for autotrophic N removal.11,20,21 In counter-

69

diffusion nitrifying MABRs, active bacteria thrive at the base of the biofilm, where they utilize oxygen

70

supplied from the membrane lumen. Growth of bacteria– including NOB- at the biofilm base would limit the

71

chance for outcompetition, once established, due to spatial protection by the overlying biofilm layers.

72

Efficient operation of MABRs to attain long-term nitritation has, to our knowledge, not been documented,

73

with the exception of one, highly-loaded (33 g-N/m2/day) fully NH4 penetrated MABR where controlling

74

DO concentrations at the membrane-biofilm interface sufficed to maintain nitritation.22

75

Recently, Pellicer-Nàcher et al.21 observed that fully nitratation MABRs accumulated NO2 immediately after

76

switching from continuous to intermittent aeration, even at elevated oxygen loadings. The causal link

77

between nitritation onset and aeration regime change were not explored. Here we report additional

78

experimental evidence of NOB suppression in intermittently aerated MABRs and we develop and calibrate

79

an improved MSNBM incorporating explicit pH calculation. Using the calibrated model, we systematically

80

evaluate potential causes for NOB suppression associated with intermittent aeration. From this analysis, we

81

identify the periodic FA inhibition- caused by transient pH upshifts and decreases at the biofilm base- as the

+

+

+

-

4 ACS Paragon Plus Environment

Page 5 of 27

Environmental Science & Technology

82

likely key cause for NOB suppression. A suitable operational window for an effective nitritation control in

83

counter-diffusion systems is finally proposed.

84

Materials and Methods

85

2.1 Reactor Operation and Measurement Methods

86

Reactor Configuration and Operation

87

The counter-diffusion MABRs consisted of two tubular gas filled PDMS membranes (3100506, Labmarket,

88

Germany), both fixed in parallel to its longer dimension (Figure S1). The system had a liquid volume of 0.83

89

L (reactors: 31.5×5×3.5 cm) and was inoculated with enriched nitrifying biomass.21 To start up the system,

90

the reactor was first run in a batch mode with an initial NH4+ concentration at 300 mg-N/L and continuous

91

aeration. The onset of NH4+ consumption without oxygen accumulation in the bulk suggested biomass

92

attachment around the membranes. Subsequently, the MABR was operated in continuous flow mode under

93

intermittent aeration. Synthetic wastewater was fed continuously with an NH4+ concentration at 75 mg-N/L

94

and without external organic carbon. Hydraulic retention time was 12 hours. The intermittent aeration

95

strategy consisted of a 6-hour aeration period (100% air) followed by a 6-hour non-aeration period (100%

96

N2). The aeration cycles were controlled by a set of solenoid valves and the pressure in the lumen was 35

97

kPa. The bulk phase was completely mixed by recirculating at 1.5 L/min. DO and pH were measured with

98

electrodes in the recirculation line (CellOX 325 and Sentix 41, WTW, Germany). Bulk pH was not

99

controlled and remained at 7.2±0.2 due to adequate buffer capacity (molar ratio in the influent: HCO3-/NH4+

100

= 2.1). Reactor temperature was at 32.5 ±0.7°C, which was above ambient temperature due to the

101

unintentional heat added by the recirculation pump. The working temperature was not controlled to a lower

102

value, as temperature effect on nitritation success in MABRs is minor.23,24 N concentrations (NH4 , NO2 and

103

NO3 ) were measured with colorimetric test kits (Spectroquant 14776, 00683, 09713; Merck, Germany).

104

Microelectrode Measurements

105

Commercially available DO microelectrode (OX-10, Unisense, Denmark) and lab-made potentiometric

106

microelectrodes for NH4 , NO2 and NO3 25 were used for in-situ profiling measurements within the biofilm.

+

-

-

+

-

-

5 ACS Paragon Plus Environment

Environmental Science & Technology

Page 6 of 27

107

Profiling measurements were performed after biofilms reaching steady state. Microelectrodes were

108

controlled by a motorized micromanipulator to a precision up to 10 µm, and began from the top of the

109

biofilm. During measurements, the influent and recirculation were kept unchanged. For each profile,

110

replicates (n > 3) were made and the average was considered in model fitting. Besides calibration following

111

the protocols, the signal drift of N-species sensors over time was corrected by measuring N concentrations

112

from effluent before and after profiling.

113

2.2 Model Development

114

The MSNBM is a one-dimensional model based on Terada et al.,26 incorporating additional explicit pH

115

calculation (Table S1). It was implemented in AQUASIM V2.1 with two compartments: a completely mixed

116

gas compartment and a biofilm compartment containing biofilm and bulk liquid.27 In the counter-diffusion

117

regime, a physical diffusion link connects the gas compartment to the base of the biofilm, defined as

118

𝐴 ∙ 𝑘𝑀,𝑖 (𝐻 𝐶𝑖,𝑎𝑖𝑟 − 𝐶𝑖,𝑏𝑎𝑠𝑒 )

119

where Ci,air and Ci,base are concentrations of carbon dioxide (CO2) or oxygen (O2) in the gas compartment and

120

at the biofilm base (mg/L), H is the non-dimensional Henry’s Law coefficient (1.32 for CO2, 34.55 for O2,

121

33 ℃), kM,i is the silicone membrane gas mass transfer coefficient (kM,O2 = 6 m/d, kM,CO2 = 0.8 m/d, Table S3).

122

Gas transfer of N2 and NH3 are not modeled. Other major modeling assumptions- regarding biofilm structure,

123

diffusion mass transfer and boundary layer thickness are as in Terada et al.26 Process rate expressions are

124

shown (Table S2). The calibrated nitrification model incorporating pH is available from the corresponding

125

author.

126

Biological Processes

127

The MSNBM includes 3 active microbial groups- AOB, NOB, heterotrophs (HB) and inerts accumulated

128

during decay processes. For the two-step nitrification process, FA and FNA are considered as true substrates

129

for growth and inhibition in nitritation and nitratation.28 The growth rate expressions are described as follows,

130

AOB: 𝜇𝐴𝑂𝐵 · 𝑋𝐴𝑂𝐵 ·

1

(1)

𝑖

𝑆𝑂2 𝐴𝑂𝐵 +𝑆 𝐾𝑂2 𝑂2

·

𝑆𝐹𝐴 𝐴𝑂𝐵 +𝑆 +𝑆 ·𝑆 ⁄𝐾 𝐴𝑂𝐵 𝐾𝐹𝐴 𝐹𝐴 𝐹𝐴 𝐹𝐴 𝐼,𝐹𝐴

·

𝐴𝑂𝐵 𝐾𝐼,𝐹𝑁𝐴 𝐴𝑂𝐵 𝐾𝐼,𝐹𝑁𝐴 +𝑆𝐹𝑁𝐴

(2)

6 ACS Paragon Plus Environment

Page 7 of 27

Environmental Science & Technology

𝑆𝑂2 𝑁𝑂𝐵 𝐾𝑂2 +𝑆𝑂2

·

𝑆𝐹𝑁𝐴 𝑁𝑂𝐵 𝑁𝑂𝐵 𝐾𝐹𝑁𝐴 +𝑆𝐹𝑁𝐴 +𝑆𝐹𝑁𝐴 ·𝑆𝐹𝑁𝐴 ⁄𝐾𝐼,𝐹𝑁𝐴

NOB: 𝜇𝑁𝑂𝐵 · 𝑋𝑁𝑂𝐵 ·

132

where μ is the specific growth rate coefficient (1/day), dependent on local pH and 𝜇𝑚𝑎𝑥 ; 𝑆𝑂2, 𝑆𝐹𝐴 and 𝑆𝐹𝑁𝐴

133

are O2, FA and FNA concentrations (mg/L), respectively; 𝐾𝑂2 , 𝐾𝐹𝐴 and 𝐾𝐹𝑁𝐴 are half-saturation coefficients

134

(mg/L);𝐾𝐼,𝐹𝐴 and 𝐾𝐼,𝐹𝑁𝐴 are inhibition coefficients (mg/L). Growth substrate inhibition (FA for AOB, FNA

135

for NOB) is incorporated with the Andrews equation. Other types of inhibition (FA for NOB, FNA for AOB)

136

are described with a noncompetitive inhibition term, as routinely done in similar studies.28, 29

137

For the denitrification process, NO2- and NO3- are modeled as separate electron acceptors. To avoid

138

unnecessary complexity and focus on AOB/NOB competition, no intermediates (NO or N2O) are considered.

139

Bacteria have different decay rates in aeration and non-aeration periods: to simplify the model, AOB/NOB

140

are assumed not to decay under anoxic or anaerobic conditions,30 meanwhile, HB decay is modified by an

141

anoxic reduction factor during non-aeration periods.

142

Chemical Process: pH Calculation

143

The one-dimensional model can keep track of local pH changes perpendicular to the membrane substratum.

144

pH along biofilm depth is calculated based on the proton production via nitrification and consumption via

145

denitrification, the equilibrium reaction with bicarbonate buffer, and CO2 stripping to the membrane lumen.

146

The consumption of inorganic carbon for autotrophic growth is neglected as it has insignificant influence on

147

pH changes under conditions when inorganic carbon is not limiting.

148

Protons produced and consumed in bioprocesses are listed in the stoichiometry matrix. The acid-base

149

balance reaction with bicarbonate buffer is assumed to occur much faster than biological processes.31

150

𝐻 + + 𝐻𝐶𝑂3− ↔ 𝐻2 𝐶𝑂3 (𝐶𝑂2 )

151

where SH, SHCO3- and SH2CO3(CO2) are concentrations of proton, bicarbonate and the sum of carbonic acid and

152

dissolved carbon dioxide, respectively (μmol/L); 𝐾𝑎,𝐻𝐶𝑂3 is the dissociation equilibrium constant of carbonic

153

acid (0.574 μmol/L, 33 ℃, 1 atm). Protons produced in the nitritation process titrate HCO3- to H2CO3, and

154

over-saturated CO2 diffuses from the biofilm base to the membrane lumen (Equation 1). Acid-base reactions

𝑆𝐻𝐶𝑂3− ·𝑆𝐻 𝐾𝑎,𝐻𝐶𝑂3

rate: (

·

𝑁𝑂𝐵 𝐾𝐼,𝐹𝐴 𝑁𝑂𝐵 𝐾𝐼,𝐹𝐴 +𝑆𝐹𝐴

131

− 𝑆𝐶𝑂2 ) · 107

(3)

(4)

7 ACS Paragon Plus Environment

Environmental Science & Technology

Page 8 of 27

155

with phosphate ions were minor and neglected, as the molar ratio of H2PO4-/HCO3- in influent was lower

156

than 3%.

157

Limitations/Inhibitions of AOB/NOB Activity

158

The growth rate expressions of AOB and NOB consider DO and pH effects. DO limitation is assessed by

159

oxygen affinity constants. Two pH effects are included. (1) pH-enzyme effect: pH can affect nitrifying

160

activity directly by changing the enzyme reaction mechanism or increasing the demand for maintenance

161

energy.31,32 A Gaussian bell-shaped curve is chosen to model the pH-enzyme dependency of specific growth

162

rates.13

163

μ=

164

where 𝜇𝑚𝑎𝑥 is the maximum specific growth rate at the optimal pH- 𝑝𝐻𝑜𝑝𝑡 , ω is the pH range within which

165

µ is larger than a half of 𝜇𝑚𝑎𝑥 . (2) pH substrate-speciation effect: local pH values determine FA/FNA

166

speciation from total NH4+/NO2-. The speciation between ionized/unionized species is assumed at

167

instantaneous equilibrium.33

168

𝑆𝐹𝐴 =

169

where 𝐾𝑎,𝑁𝐻3 and 𝐾𝑎,𝑁𝑂2 are dissociation equilibrium constants of ammonium and nitrous acid, respectively

170

(0.000794 and 628.96 μmol/L (33 ℃, 1 atm)). Substrate-speciation will result in differential degrees of

171

FA/FNA inhibition.

172

2.3 Sensitivity Analysis and Parameter Estimation

173

To investigate the most determinant parameters on reactor performance, a sensitivity analysis was performed.

174

Initial values of kinetic parameters were taken from ASMN model.28 The optimal pH ranges for AOB and

175

NOB growth kinetics (pHopt and ω) were from Park et al..13 The temperature correction for µmax and bmax are

176

from Hao et al..34 The MSNBM was first run in continuous aeration with default values for 300 days to

177

achieve a stable nitrifying biofilm. Then a local sensitivity analysis was performed after switching to

178

intermittent aeration- giving individual parameter a 100% value change while all others remained constant.27

𝜇𝑚𝑎𝑥 2

𝜋

{1 + 𝑐𝑜𝑠 [𝜔 · (𝑝𝐻 − 𝑝𝐻𝑜𝑝𝑡 )]}

𝐾𝑎,𝑁𝐻3 ·𝑆𝑁𝐻4 𝑆𝐻

𝑆𝐹𝑁𝐴 =

|𝑝𝐻 − 𝑝𝐻𝑜𝑝𝑡 | < 𝜔

𝑆𝑁𝑂2 ·𝑆𝐻 𝐾𝑎,𝑁𝑂2

(5)

(6)

8 ACS Paragon Plus Environment

Page 9 of 27

Environmental Science & Technology

𝑆𝑁𝐻4,𝑖𝑛 −𝑆𝑁𝐻4,𝑒𝑓𝑓

%),

179

Reactor performances were evaluated in terms of ammonium removal efficiency (ARE,

180

nitrate production efficiency (NaE,

181

NOB fraction (fNOB, 𝑁𝑂𝐵+𝐴𝑂𝐵 %). The normalized sensitivity function is defined as,

182

2 𝛿𝑗 = √𝑎𝑣𝑒𝑟𝑎𝑔𝑒(𝑆𝑒𝑛𝑠𝑖,𝑗 ) 𝑎𝑛𝑑 𝑆𝑒𝑛𝑠𝑖,𝑗 = 𝑝𝑖,𝑗 ∆𝑝 ,

183

where 𝛿𝑗 , 𝑦𝑗 and pi,j are the sensitivity function, the output reactor performances (ARE, NaE, NE or fNOB),

184

and the input parameters, respectively. Sensi,j was evaluated at different times during the aeration cycles

185

(time interval of 0.01 day) and at 20 equidistant points within the biofilm or 1 point in the bulk phase. The

186

averaged value was considered in the sensitivity analysis and parameter sensitivity was ranked for each

187

targeted performance metric. We focused on biokinetic and stoichiometric parameters related to AOB and

188

NOB, as HB parameters are of secondary importance in nitrifying biofilms.35

189

The most sensitive parameters were calibrated with steady state experimental data. The model calibration

190

was carried out by trial and error through adjusting the parameter values one by one to minimize the fitting

191

error. Root mean squared error was used to assess the quality of model-data fit as the objective function,

192

RMSE = √𝑎𝑣𝑒𝑟𝑎𝑔𝑒(∑𝑗 ∑𝑖(

193

where j is the targeted variable measured or estimated (NH4 , NO2 , NO3 and DO), i is a sample point along

194

biofilm depth (i =20). The model was validated with additional experimental data from this MABR and

195

experimental data from a separate membrane-aerated biofilm reactor (MABR2) operated under 4 different

196

ammonium surface loadings (Table S5, detailed description of the experimental data used in model

197

calibration and validation).36 The calibrated parameters were checked by comparing RMSE in the calibration

198

with RMSE in the validation and the Janus coefficient (J) was calculated,37

199

𝐽2 =

200

2.4 Model Simulations

201

The calibrated MSNBM was run in 3 scenarios (Table S6, detailed description of each simulation scenario):

𝑆𝑁𝑂3,𝑒𝑓𝑓 𝑆𝑁𝐻4,𝑖𝑛 −𝑆𝑁𝐻4,𝑒𝑓𝑓

%), nitritation efficiency (NE,

𝑆𝑁𝐻4,𝑖𝑛 𝑆𝑁𝑂2,𝑒𝑓𝑓

𝑆𝑁𝐻4,𝑖𝑛 −𝑆𝑁𝐻4,𝑒𝑓𝑓

%) and

𝑁𝑂𝐵

∆𝑦𝑗

(7)

𝑖,𝑗

𝑦𝑚𝑜𝑑𝑒𝑙,𝑖,𝑗 −𝑦𝑚𝑒𝑎𝑠,𝑖,𝑗 2 ) ) 𝑦𝑚𝑒𝑎𝑠,𝑗,𝑎𝑣𝑒𝑟𝑎𝑔𝑒

(8)

+

-

𝑅𝑀𝑆𝐸𝑣𝑎𝑙 2 𝑅𝑀𝑆𝐸𝑐𝑎𝑙 2

-

(9)

9 ACS Paragon Plus Environment

Environmental Science & Technology

Page 10 of 27

202

(1) To validate the model with extra experimental data, the calibrated MSNBM was ran in intermittent

203

aeration (6-hour aeration period and 6-hour non-aeration period) under different NH4 surface loadings or in

204

continuous aeration in a batch test. Then the determinant factor(s) that govern NOB suppression in this

205

MABR was explored with the validated model.

206

(2) To clarify why NOB suppression occurred after switching to intermittent aeration from continuous

207

aeration, the model was run in continuous aeration to achieve a nitrifying biofilm, then aeration was switched

208

to the same intermittent aeration as scenario 1.

209

(3) To optimize the operational window for nitritation in intermittently aerated MABRs, different

210

intermittent aeration strategies and influent concentrations were simulated in MSNBM after achieving a

211

nitrifying biofilm in continuous aeration. The effects of aeration intermittency and residual NH4 (FA)

212

concentrations on NOB suppression were evaluated.

213

Results and Discussion

214

3.1 Model Calibration and Evaluation

215

A sensitivity function, considering the sum of reactor performances (ARE, NaE, NE and fNOB), was

216

𝐴𝑂𝐵 𝐴𝑂𝐵 𝑁𝑂𝐵 , followed by 𝐾𝐼,𝐹𝐴 , 𝜇𝑚𝑎𝑥 , calculated to rank parameters (Figure S2). The most sensitive parameter is 𝜇𝑚𝑎𝑥

217

𝑁𝑂𝐵 𝐴𝑂𝐵 𝑁𝑂𝐵 𝐾𝐼,𝐹𝐴 , 𝐾𝑂2 and 𝐾𝑂2 . The ranking shows that µmax is the most determinant among all kinetic parameters in

218

nitrogen conversion simulations. It is consistent with the sensitivity analysis of Wang et al.35 who ranked

219

kinetic parameters in terms of nitritation performance and biofilm development in nitrifying biofilm reactors.

220

The higher sensitivity regarding performance within the biofilm (Figure S2B) versus the bulk (Figure S2A)

221

suggests that in-situ microprofiling data is more informative in model calibration than bulk measurements,

222

which were typically used.22,38 Therefore, microprofiling measurements (NH4+, NO2-, NO3- and DO) in the

223

first aeration hour at steady state were used to calibrate sensitive parameter(s). Microprofiles in the last

224

aeration hour (NH4+, NO2-, NO3- and DO) and bulk profiles during an intermittent aeration cycle (NH4+, NO2-,

225

NO3-, DO and pH) representing the reactor performance at steady-state were used for validation. Additional

226

validation of the model and its parameter estimates was obtained by fitting the initial reactor performance

+

+

10 ACS Paragon Plus Environment

Page 11 of 27

Environmental Science & Technology

-

+

-

+

227

(NH4 , NO2 and NO3 ) when operated in batch start-up mode, and by fitting the biofilm performance (NH4 ,

228

NO2 , NO3 and pH) of a separately operated MABR under different NH4 surface loadings.

229

𝐴𝑂𝐵 By fitting the most sensitive parameter- 𝜇𝑚𝑎𝑥 in the reported range,12 the RMSE decreased to 0.5 and the

230

𝑁𝑂𝐵 deviation in NO3 fitting contributed the most to the error. Thus, the next most sensitive parameter- 𝜇𝑚𝑎𝑥 - in

231

NO3 sensitivity ranking (Figure S3) was added to the calibration and RMSE decreased to 0.1. Values of

232

𝐴𝑂𝐵 𝑁𝑂𝐵 𝜇𝑚𝑎𝑥 and 𝜇𝑚𝑎𝑥 were within a reasonable range: the estimated maximum growth rates at the optimal pH were

233

2.35 d-1 for AOB and 2.15 d-1 for NOB (Table 1). In the data fitting, the error function was bounded, but

234

𝑁𝑂𝐵 𝐴𝑂𝐵 and 𝜇𝑚𝑎𝑥 were highly correlated indicating a poorly identifiable parameter set. Model evaluations 𝜇𝑚𝑎𝑥

235

𝐴𝑂𝐵 𝑁𝑂𝐵 were, however, not affected by changes in the 𝜇𝑚𝑎𝑥 (2.35-2.85) and 𝜇𝑚𝑎𝑥 (2.15-2.55) best-fit parameter

236

𝐴𝑂𝐵 𝑁𝑂𝐵 value region (Figure S3D, approximate >99% confidence region: 𝜇𝑚𝑎𝑥 2.25-2.95 and 𝜇𝑚𝑎𝑥 2.06-2.66).

237

Predicted microprofiles agree with measurements in the first aeration hour at steady state (Figure 1A): NH4

238

is consumed along biofilm depth and NO2 is produced; NO3 remains at lower concentrations than NO2

239

within the biofilm; DO penetrates 60 µm into the biofilm base. The greatest divergence in the overall fitting

240

corresponds to NO2 at the biofilm base (6 mg-N/L) but only overestimates FNA concentrations by 0.002

241

mg-N/L. Errors in DO fitting at the membrane-biofilm interface (6.6 mg/L predicted versus 1.7 mg/L

242

measured) have a minor influence on the oxygen competition between AOB and NOB (Table S4), consistent

243

with Lackner and Smets39 who reported that oxygen concentrations at interfaces were not decisive in

244

nitritation performance in MABRs. Additionally, uncertainty in measuring the interface DO could be caused

245

by microbial activities on the membrane and an efficiency factor E (1.3 ~ 4.3) was suggested to correct

246

measured values.36

247

MSNBM predicts consistent profiles in the different model validations. It predicted lower NH4+ and higher

248

NO2- within the biofilm in the last aeration hour (Figure 1B) and uniform dynamic variations of bulk

249

concentrations during a 12-hour intermittent aeration cycle. For example, it captured the pH decreases in the

250

6-hour aeration phase and increases in the 6-hour non-aeration phase (Figure 1C). It also predicted

251

simultaneous production of NO2- and NO3- in the batch mode data validation (Figure S4A) and predicted

252

NH4+ consumption and NO2- production following the tendencies observed in MABR2 (Figure S4B). Janus

-

-

+

-

-

-

-

+

-

-

11 ACS Paragon Plus Environment

Environmental Science & Technology

Page 12 of 27

253

coefficients were around 1.9 (± 0.5), showing that the RMSEs were within the same order of magnitude in

254

calibration and validations.

255

3.2 Model-based Exploration of NOB Suppression in Intermittently Aerated MABRs

256

NOB suppression is the result of indirect and direct (competitive) interactions between AOB and NOB in the

257

local environment. Net microbial activities are captured in the specific growth rates: biomass types with the

258

higher specific growth rate will win the local competition. In the studied system, oxygen was provided

259

intermittently from membrane lumen. The biomass type with the higher specific growth rate (AOB or NOB)

260

thus dominated the oxygen utilization.

261

Consistent with experimental reactor operations, simulations were initiated with fully-nitrifying biomass and

262

subject to intermittent aeration. Both simulation and experimental data showed that after 2 weeks in

263

intermittent aeration bulk N concentrations became stable, especially NO3 was below 1 mg-N/L indicating

264

efficient suppression of NOB activity (Figure S9). To illustrate the competition in the first nitrifying stage,

265

profiles of specific growth rates of AOB and NOB during an aeration cycle (6 hours) are plotted at day 15

266

(Figure 2A). The averaged µ at time intervals shows kinetic variations over time: (1) 0-15 minutes, with the

267

onset of aeration microbial activities recover from the previous non-aeration period and increase

268

dramatically; (2) 15-180 minutes, AOB activity becomes stable, while NOB activity still recovers; (3) 180-

269

360 minutes, both AOB and NOB activity reach pseudo steady state. The model shows the ratio of 𝜇𝐴𝑂𝐵 to

270

𝐴𝑂𝐵 𝑁𝑂𝐵 𝜇𝑁𝑂𝐵 increases in the intermittent aeration, compared to the ratio of 𝜇𝑚𝑎𝑥 to 𝜇𝑚𝑎𝑥 in continuous aeration

271

(1.5±0.15 versus 1.1). AOB preferentially utilize oxygen to support growth while NOB are outcompeted or

272

their activity is suppressed.

273

To assess the relative contribution of DO/pH effects on NOB suppression, individual factors influencing

274

growth rates were calculated spatially (at different biofilm depths) and temporally (at different times during

275

the cycle). Considering the effective DO penetration depth, only results in the first 100-μm at the biofilm

276

base are shown (Figure 2B).

-

12 ACS Paragon Plus Environment

Page 13 of 27

Environmental Science & Technology

277

DO Limitation in NOB Suppression

278

O2 is a growth substrate for both AOB and NOB. In counter-diffusion biofilms O2 is provided via the lumen

279

and NH4 via the bulk. In the biofilm, DO penetrates only 60 μm during aeration periods with the highest

280

concentration at the membrane-biofilm interface (biofilm depth= 0 µm), presenting spatial variations (Figure

281

S5A). Besides, DO varies over time during aeration cycles. DO at the membrane-biofilm interface is 0 mg/L

282

at the onset of aeration and quickly increases to the maximum concentration within 15 minutes. Afterwards,

283

DO concentrations within the biofilm remain stable until the end of aeration.

284

The DO limitation effect was evaluated based on oxygen concentrations within the biofilm (Figure 2B, 1-

285

DO limitation). In aeration periods, during the first 15 minutes DO strongly limits both AOB and NOB

286

activities. During the following period, the limitation is alleviated as DO increases and stabilizes, but still

287

remains strong above 30 µm. With a lower DO affinity NOB are more oxygen-limited than AOB. However,

288

the relatively stronger limitation to NOB is insignificant in its suppression. Model results show that oxygen

289

transfer and its diffusion mostly affects NH4 oxidation efficiency rather than nitritation efficiency (Table

290

S7).

291

pH-enzyme Effect on NOB Suppression

292

Because pH affects AOB/NOB kinetics directly and indirectly, it is necessary to incorporate pH effects in

293

models.14,40 Here MSNBM predicts local pH values within the biofilm and the response to transient aeration

294

phases (Figure S5B). While measurements showed that bulk pH remained relatively stable (±0.2), pH within

295

the biofilm, especially in the DO-penetrated zone, showed considerable variations (±0.6). At the onset of

296

aeration the model indicates a transient pH upshift at the biofilm base (0-15 minutes). The accumulated

297

alkalinity is attributed to continuous CO2 diffusion from the biofilm base to the membrane lumen where N2

298

gas flows through in the previous non-aeration period and slight denitrification activities. As aeration

299

continues, pH decreases due to proton production associated with NH4 oxidation. Simulations predict that

300

pH within the biofilm becomes lower than in the bulk after 1-hour aeration and decreases slowly afterwards.

301

At the end of aeration pH at the biofilm base is 0.4 units lower than the average bulk pH, which will increase

+

+

+

13 ACS Paragon Plus Environment

Environmental Science & Technology

Page 14 of 27

302

again in the following non-aerated phase. Thus pH varies periodically in the intermittently aerated biofilms,

303

a pattern similar but slower than DO variations.

304

The pH-enzyme effect was assessed based on local pH values (Figure 2B, 2- pH-enzyme effect). It favors

305

NOB growth over AOB as NOB have a lower pHopt (NOB: 7.7 versus AOB: 8.4) and pH varies in the

306

optimal range for its growth. Moreover, the pH-enzyme effect is also insignificant in the overall AOB/NOB

307

competition due to their robust growth in broad pH ranges and the relatively small pH variations in the

308

system.

309

pH Substrate-speciation Effects on NOB Suppression

310

FA/FNA concentrations rely on pH values as well as total NH4 /NO2 concentrations. In counter-diffusion

311

biofilms, NH4 , provided via the bulk, is oxidized at the biofilm base producing NO2 which diffuses

312

backward into the bulk.10 Based on ionic N concentrations, FA and FNA speciation synchronizes with pH

313

variations (Figure S5C and S5D). For instance, at the onset of aeration FA concentration is high due to NH4

314

and alkalinity accumulation from the previous non-aeration period. During the following aeration period, FA

315

concentration decreases, as pH drops and NH4 consumption continues. On the other hand, FNA shows

316

reversed variations: increasing as aeration progresses and with biofilm depth as a result of the proton and

317

NO2 production.

318

The pH substrate-speciation effect was assessed based on FA/FNA concentrations within the biofilm (Figure

319

2B, 3- FA/FNA inhibition). During the first 15 minutes, FA strongly inhibits AOB/NOB microbial activities

320

(FA > KI,FA). Afterwards, the inhibition is alleviated as FA decreases. Noticeably, FA inhibits AOB and

321

NOB in different ways: the inhibition effect remains strong for NOB throughout the aeration period (from

322

0.26 to 0.62), while it obviously weakens for AOB (from 0.54 to 0.89). FA inhibition rapidly becomes the

323

most determinant factor in suppressing NOB over AOB. As FNA concentrations are always an order of

324

magnitude lower than KI,FNA, its inhibition effect on microbial activities is always minor thereby contributing

325

little to NOB suppression.

+

-

-

+

+

+

-

14 ACS Paragon Plus Environment

Page 15 of 27

Environmental Science & Technology

326

Besides the inhibitor effect (𝐾𝐼 /(𝐾𝐼 + 𝑆)), FA/FNA exhibit the substrate limitation effect (𝑆/(𝐾𝑆 + 𝑆)) in

327

biological processes (Equation 2). However, FA and FNA concentrations are far above the substrate

328

𝐴𝑂𝐵 𝑁𝑂𝐵 affinities (𝐾𝐹𝐴 and 𝐾𝐹𝑁𝐴 ) in the system, making the substrate limitation effects negligible.

329

Implication of Model-based NOB Suppression

330

Overall, FA inhibition caused by pH substrate-speciation is the crucial factor in suppressing NOB in the

331

intermittently aerated biofilm reactors. Nitritation success is insensitive to oxygen affinity constants or DO

332

concentrations at the membrane-biofilm interface- a conclusion different from previous studies.41,42 Downing

333

and Nerenberg22 suggested manipulating interface DO as an effective method to control shortcut nitrification

334

in MABRs: with a lower interface DO, more NO2 accumulated. However, their biofilms performed at low

335

nitrification rates with a low influent NH4 concentration- 3 mg-N/L, suggesting little FA inhibition and no

336

NO2 accumulation or significant pH gradients. The single DO gradient within the biofilm present the

337

interface DO as a key role in nitritation success. This method might not apply for N-rich wastewater

338

treatment. For example Lackner and Smets39 concluded that nitritation success based only on interface DO

339

was not possible in a counter-diffusion biofilm with high influent NH4 concentrations (20-800 mg-N/L), and

340

nitritation efficiency was not predicted from oxygen affinity constants.

341

Counter- and co-diffusion biofilms have different mechanisms of NOB suppression due to different spatial

342

structures and population distributions.35,38,39 In counter-diffusion biofilms, the theoretically optimal habitat

343

for NOB is the biofilm base, where both SO2/KO2 and SFNA/KFNA have the highest values. By contrast, the

344

base is not the optimal for AOB growth, as SO2/KO2 and SFA/KFA cannot have the maximum at the same

345

spatial position. Outcompeting NOB can be more difficult in counter-diffusion over co-diffusion biofilms,

346

where microbes (AOB and NOB) share the optimal habitats at the biofilm top near the biofilm/liquid

347

interphase. Others have similarly observed that NOB could survive better in counter- versus co- diffusion

348

biofilms, even when operated under constant oxygen limited (DO < 0.1 mg/L) and high pH (8.0-8.3)

349

conditions in the bulk.35 The inherent system geometry of membrane-aerated biofilms complicates NOB

350

inhibition/washout. Besides, when applying intermittent aeration, periodic pH variations at the biofilm base

351

exert a significant effect on NOB dynamics in counter-diffusion biofilms because of continuous CO2

-

+

-

+

15 ACS Paragon Plus Environment

Environmental Science & Technology

Page 16 of 27

352

diffusion to the gas lumen. However, such pH variations are not expected in co-diffusion biofilms. Many

353

studies have highlighted the benefits of low DO with high FA to maintain shortcut NH4 removal in co-

354

diffusion biofilms.17,43 Park et al.3 explored simultaneous effects of DO and FA/FNA in lab-scale co-

355

diffusion nitrifying biofilms, and found that NO2 accumulated due to DO limitation or FA inhibition and

356

long-term NOB suppression could not be maintained without DO limitation involved. The results were

357

consistent with Brockmann and Morgenroth44 who suggested that oxygen limitation was the main

358

mechanism for NOB suppression and FA inhibition was not necessarily required in co-diffusion biofilms.

359

However, DO limitation in nitritation counter-diffusion biofilms appears not as significant as reported for co-

360

diffusion biofilms, consistent with the observation that nitritation could not be achieved by solely

361

manipulating air pressure in the membrane lumen in MABRs.21

362

3.3 Potential explanation of NOB Suppression in the study of Pellicer-Nàcher et al. (2010)

363

To answer why NO2 accumulated after switching from continuous to intermittent aeration in MABRs,

364

simulations were carried out with the calibrated MSNBM in continuous aeration for 200 days followed by

365

intermittent aeration (6-hour aeration and 6-hour non-aeration cycles). The simulation shows a nitrifying

366

biofilm during continuous aeration (NE = 0%) indicating no NOB suppression (Table 2- continuous

367

aeration). After switching to intermittent aeration the model predicts NOB suppression- NO3 decrease and

368

NE increase (Table 2- strategy A, Figure S6). To find the critical factor for NOB suppression, variations of

369

individual pH/DO effect on AOB/NOB competition were assessed: each effect

370

aeration (for instance at day 215) was normalized by its value during continuous aeration. A value higher

371

than 1 means the effect favors NOB suppression in intermittent aeration, and lower than 1 that it favors NOB

372

growth.

373

Only FA inhibition is identified to favor NO2 accumulation after switching the aeration strategy, while DO

374

limitation, pH-enzyme effect and FNA inhibition remain unchanged (Figure S7). FA inhibition shows certain

375

varying patterns in intermittent aeration: (1) it is overall enhanced due to an increased residual NH4 ; (2) it is

376

particularly strong during the first 15 minutes of aeration. The simulated increase of residual NH4+ after

377

changing to intermittent aeration was also observed in the study of Pellicer-Nàcher et al.:21 in reactor B bulk

+

-

-

-

𝑒𝑓𝑓𝑒𝑐𝑡𝐴𝑂𝐵 𝑒𝑓𝑓𝑒𝑐𝑡𝑁𝑂𝐵

in intermittent

-

+

16 ACS Paragon Plus Environment

Page 17 of 27

Environmental Science & Technology

378

NH4+ increased by 100 mg/L at stage 1 and 2 (intermittent aeration) compared to stage 0 (continuous

379

aeration). Compared to continuous aeration, MABRs in intermittent aeration display a tradeoff between

380

NH4 removal efficiency and nitritation efficiency (Table 2). Nitritation is assisted by the evaluated residual

381

NH4 , which underlines the importance of a minimum NH4 concentration in the bulk. Pérez et al.18 also

382

highlighted the need for minimum residual NH4 for NOB suppression in co-diffusion biofilms, but

383

attributed the nitritation success to differential oxygen limitation rather than FA inhibition- as NOB were

384

outcompeted due to the strong oxygen limiting conditions imposed by a high residual NH 4 . The strong FA

385

inhibition at the onset of aeration is due to pH upshifts at the biofilm base in the previous anoxic phases. It

386

causes a longer lag phase of NOB activity over AOB, which could be another reason in the nitritation

387

success. Theoretically, NOB locate at the biofilm base, if enriched in MABRs, thus pH upshift at the base is

388

more efficient to prompt FA inhibition than increasing bulk pH. This lag phase has also been observed in

389

other intermittently aerated systems.45,46 Kornaros et al.47 and Gilbert et al.48 attributed the lag phase to a long

390

(enzyme) reactivation time in NOB nitrogen metabolism after anoxic exposure in batch continuous stirred-

391

tank reactors. However, the possibility for pH variations was not considered in those studies, even though

392

CO2 stripping could slowly increase bulk pH.49

393

3.4 Nitritation in Various Intermittent Aeration Strategies

394

For an intermittent aeration system with certain NH4 /O2 surface loadings, the aeration duration determines

395

residual NH4 concentrations: a longer aeration lowers residual NH4 . The aeration intermittency determines

396

pH upshift times and the variation range of bulk concentrations: a higher frequency causes more pH upshifts

397

and a narrow variation range. This information can be utilized to optimize intermittent aeration strategies for

398

efficient nitritation in MABRs (Table 2). MSNBM simulation shows that a higher aeration intermittency can

399

accelerate NOB suppression (A and C) due to more times of pH upshift in non-aeration phases to retard

400

NOB activity while slightly affecting AOB activity, or decelerate NOB suppression (B and A) due to the

401

relatively high bulk NH4 (pH) at the onset of aeration phases even the averaged bulk concentrations are the

402

same. Longer aeration duration (D) leads to a slower nitritation process but a higher NH4 removal efficiency,

403

while keeping the same aeration intermittency. It is consistent with the observation in Mota et al.47 that

+

+

+

+

+

+

+

+

+

+

17 ACS Paragon Plus Environment

Environmental Science & Technology

Page 18 of 27

404

intermittently aerated reactors with longer anoxic phase had the lower NOB abundance and relatively higher

405

NH4 effluent concentrations. Both studies suggest that the maximum aeration duration should be set to

406

ensure nitritation success in intermittent aeration, and a specific to the treated wastewater ratio of aeration to

407

non-aeration phase is needed to balance NOB suppression against NH4 removal.51 Simulation with high

408

NH4+ concentrations predicts fast nitritation in the intermittent aeration (E), and vice versa slow nitritation

409

with low influent NH4+ (F). Further simulation with low NH4+ concentrations but high bulk pH (G) shows

410

efficient nitritation, confirming a key factor in NOB suppression was bulk FA rather than residual NH 4+

411

(more simulations in Table S8). In an intermittent aeration regime, the bulk FA can provide a rapid indicator

412

of the nitritation potential of MABRs (Figure S10). MSNBM simulations reveal that aeration duration and

413

intermittency control the performance of intermittently aerated nitrifying biofilms: longer aeration duration

414

ensures a higher NH4+ removal efficiency, yet impedes NOB suppression; higher aeration intermittency

415

presents unchanged NH4+ removal performance, while its effect on NOB suppression should be evaluated

416

under specific conditions. Following this model-based analysis, experimental validation of model predictions

417

is warranted.

418

In conclusion, we provide experimental evidence that intermittent aeration supports efficient nitritation in

419

membrane aerated biofilm reactors (MABRs). A pH-explicit 1-D multispecies nitrifying biofilm model

420

(MSNBM) is developed and calibrated: model analysis reveals that NOB suppression - associated with

421

intermittent aeration - is primarily governed by periodic FA inhibition as the consequence of transient pH

422

upshifts during non-aeration. These pH upshifts are mainly caused by alkalinity increases due to CO2

423

stripping to the membrane lumen (which also occurs during aeration) plus the cessation of proton production

424

(which only occurs during aeration). In counter diffusion biofilms pH effect is more important than DO

425

(limitation) effect on NOB suppression. Both aeration intermittency and duration are effective control factors

426

to obtain nitritation success in intermittently membrane-aerated biofilms, and maintaining nitritation and

427

NH4 removal efficiency is more easily ensured if operated with high buffer capacities.

+

+

+

18 ACS Paragon Plus Environment

Page 19 of 27

Environmental Science & Technology

428

Acknowledgement

429

The authors would like to thank the China Scholarship Council (CSC) for financial support to Yunjie Ma,

430

and the Innovation Fund Denmark (IFD) (Project LaGAS, File No. 0603-00523B) for additional financial

431

support.

432

Supporting Information. More detailed information on model description, model simulations and data

433

analysis. This material is available free of charge via the Internet at http://pubs.acs.org.

434

19 ACS Paragon Plus Environment

Environmental Science & Technology

Page 20 of 27

435

References

436 437 438

(1)

Hellinga, C.; Schellen, A.; Mulder, J.; Vanloosdrecht, M.; Heijnen, J. The sharon process: An innovative method for nitrogen removal from ammonium-rich waste water. Water Sci. Technol. 1998, 37 (9), 135–142.

439 440

(2)

Jenicek, P.; Svehia, P.; Zabranska, J.; Dohanyos, M. Factors affecting nitrogen removal by nitritation/denitritation. Water Sci. Technol. 2004, 49 (5–6), 73–79.

441 442 443

(3)

Park, S.; Chung, J.; Rittmann, B. E.; Bae, W. Nitrite accumulation from simultaneous free-ammonia and free-nitrous-acid inhibition and oxygen limitation in a continuous-flow biofilm reactor. Biotechnol. Bioeng. 2015, 112 (1), 43–52.

444 445

(4)

van Dongen, U.; Jetten, M. S.; van Loosdrecht, M. C. The SHARON-Anammox process for treatment of ammonium rich wastewater. Water Sci. Technol. 2001, 44 (1), 153–160.

446 447

(5)

Kuenen, J. G. Anammox bacteria: from discovery to application. Nat. Rev. Microbiol. 2008, 6 (4), 320–326.

448 449 450

(6)

Sliekers, A. O.; Haaijer, S. C. M.; Stafsnes, M. H.; Kuenen, J. G.; Jetten, M. S. M. Competition and coexistence of aerobic ammonium- and nitrite-oxidizing bacteria at low oxygen concentrations. Appl. Microbiol. Biotechnol. 2005, 68 (6), 808–817.

451 452 453

(7)

Vadivelu, V. M.; Yuan, Z.; Fux, C.; Keller, J. The inhibitory effects of free nitrous acid on the energy generation and growth processes of an enriched Nitrobacter culture. Environ. Sci. Technol. 2006, 40 (14), 4442–4448.

454 455 456

(8)

Terada, A.; Lackner, S.; Kristensen, K.; Smets, B. F. Inoculum effects on community composition and nitritation performance of autotrophic nitrifying biofilm reactors with counter‐diffusion geometry. Environ. Microbiol. 2010, 12 (10), 2858–2872.

457 458 459

(9)

Fux, C.; Huang, D.; Monti, A.; Siegrist, H. Difficulties in maintaining long-term partial nitritation of ammonium-rich sludge digester liquids in a moving-bed biofilm reactor (MBBR). Water Sci. Technol. 2004, 49 (11–12), 53–60.

460 461

(10)

Schramm, A.; De Beer, D.; Gieseke, A.; Amann, R. Microenvironments and distribution of nitrifying bacteria in a membrane-bound biofilm. Environ. Microbiol. 2000, 2 (6), 680–686.

462 463

(11)

Shanahan, J. W.; Semmens, M. J. Alkalinity and pH effects on nitrification in a membrane aerated bioreactor: An experimental and model analysis. Water Res. 2015, 74, 10–22.

464 465

(12)

Vannecke, T. P. W.; Volcke, E. I. P. Modelling microbial competition in nitrifying biofilm reactors. Biotechnol. Bioeng. 2015, 112 (12), 2550–2561.

466 467

(13)

Park, S.; Bae, W.; Chung, J.; Baek, S.-C. Empirical model of the pH dependence of the maximum specific nitrification rate. Process Biochem. 2007, 42 (12), 1671–1676.

468 469

(14)

Fumasoli, A.; Morgenroth, E.; Udert, K. M. Modeling the low pH limit of Nitrosomonas eutropha in high-strength nitrogen wastewaters. Water Res. 2015, 83, 161–170.

470 471 472

(15)

Carrera, J.; Jubany, I.; Carvallo, L.; Chamy, R.; Lafuente, J. Kinetic models for nitrification inhibition by ammonium and nitrite in a suspended and an immobilised biomass systems. Process Biochem. 2004, 39 (9), 1159–1165.

473 474 475

(16)

Martin, K. J.; Picioreanu, C.; Nerenberg, R. Assessing microbial competition in a hydrogen-based membrane biofilm reactor (MBfR) using multidimensional modeling. Biotechnol. Bioeng. 2015, 112 (9), 1843–1853.

476 477 478

(17)

Park, S.; Bae, W.; Rittmann, B. E. Multi-Species Nitrifying Biofilm Model (MSNBM) including free ammonia and free nitrous acid inhibition and oxygen limitation. Biotechnol. Bioeng. 2010, 105 (6), 1115–1130.

20 ACS Paragon Plus Environment

Page 21 of 27

Environmental Science & Technology

479 480 481

(18)

Pérez, J.; Lotti, T.; Kleerebezem, R.; Picioreanu, C.; van Loosdrecht, M. C. M. Outcompeting nitriteoxidizing bacteria in single-stage nitrogen removal in sewage treatment plants: A model-based study. Water Res. 2014, 66, 208–218.

482 483

(19)

Casey, E.; Glennon, B.; Hamer, G. Review of membrane aerated biofilm reactors. Resour. Conserv. Recycl. 1999, 27 (1–2), 203–215.

484 485

(20)

Terada, A.; Yamamoto, T.; Igarashi, R.; Tsuneda, S.; Hirata, A. Feasibility of a membrane-aerated biofilm reactor to achieve controllable nitrification. Biochem. Eng. J. 2006, 28 (2), 123–130.

486 487 488

(21)

Pellicer-Nàcher, C.; Sun, S.; Lackner, S.; Terada, A.; Schreiber, F.; Zhou, Q.; Smets, B. F. Sequential aeration of membrane-aerated biofilm reactors for high-rate autotrophic nitrogen removal: experimental demonstration. Environ. Sci. Technol. 2010, 44 (19), 7628–7634.

489 490

(22)

Downing, L. S.; Nerenberg, R. Effect of oxygen gradients on the activity and microbial community structure of a nitrifying, membrane-aerated biofilm. Biotechnol. Bioeng. 2008, 101 (6), 1193–1204.

491 492 493

(23)

Pellicer-Nàcher, C.; Sun, S.; Lackner, S.; Terada, A.; Schreiber, F.; Zhou, Q.; Smets, B. F. Sequential aeration of membrane-aerated biofilm reactors for high-rate autotrophic nitrogen removal: experimental demonstration. Environ. Sci. Technol. 2010, 44 (19), 7628–7634.

494 495

(24)

Zhu, S.; Chen, S. The impact of temperature on nitrification rate in fixed film biofilters. Aquac. Eng. 2002, 26 (4), 221–237.

496 497

(25)

Gieseke, A.; Beer, D. De. Use of microelectrodes to measure in situ microbial activities in biofilms, sediments, and microbial mats. Mol. Microb. Ecol. Man. 2004, 2, 1281–1612.

498 499 500

(26)

Terada, A.; Lackner, S.; Tsuneda, S.; Smets, B. F. Redox-stratification controlled biofilm (ReSCoBi) for completely autotrophic nitrogen removal: the effect of co- versus counter-diffusion on reactor performance. Biotechnol. Bioeng. 2007, 97 (1), 40–51.

501 502

(27)

Reichert, P. AQUASIM 2.0—Computer program for the identification and simulation of aquatic systems. Dubendorf, Switz. EAWAG. 1998.

503 504

(28)

Hiatt, W. C.; Grady, C. P. L. An updated process model for carbon oxidation, nitrification, and denitrification. Water Environ. Res. 2008, 80, 2145–2156.

505 506 507

(29)

Hellinga, C.; Van Loosdrecht, M. C. M.; Heijnen, J. J. Model Based Design of a Novel Process for Nitrogen Removal from Concentrated Flows. Mathematical and Computer Modelling of Dynamical Systems. 1999, pp 351–371.

508 509

(30)

Blackburne, R.; Yuan, Z.; Keller, J. Demonstration of nitrogen removal via nitrite in a sequencing batch reactor treating domestic wastewater. Water Research. 2008, pp 2166–2176.

510 511 512

(31)

Sötemann, S.; Musvoto, E.; Wentzel, M.; Ekama, G. Integrated biological, chemical and physical processes kinetic modelling
Part 1 – Anoxic-aerobic C and N removal in the activated sludge system. Water SA 2006, 31 (4), 1044–1062.

513 514 515

(32)

Van Hulle, S. W. H.; Volcke, E. I. P.; Teruel, J. L.; Donckels, B.; van Loosdrecht, M. C. M.; Vanrolleghem, P. A. Influence of temperature and pH on the kinetics of the Sharon nitritation process. J. Chem. Technol. Biotechnol. 2007, 82 (5), 471–480.

516 517 518

(33)

Musvoto, E. V.; Wentzel, M. C.; Loewenthal, R. E.; Ekama, G. A. Integrated chemical-physical processes modelling - I. Development of a kinetic-based model for mixed weak acid/base systems. Water Res. 2000, 34 (6), 1857–1867.

519 520 521

(34)

Hao, X.; Heijnen, J. J.; Van Loosdrecht, M. C. . Model-based evaluation of temperature and inflow variations on a partial nitrification–ANAMMOX biofilm process. Water Res. 2002, 36 (19), 4839– 4849.

522 523 524

(35)

Wang, R.; Terada, A.; Lackner, S.; Smets, B. F.; Henze, M.; Xia, S.; Zhao, J. Nitritation performance and biofilm development of co-and counter-diffusion biofilm reactors: modeling and experimental comparison. Water Res. 2009, 43 (10), 2699–2709.

21 ACS Paragon Plus Environment

Environmental Science & Technology

Page 22 of 27

525 526

(36)

Pellicer-Nàcher, C.; Domingo-Félez, C.; Lackner, S.; Smets, B. F. Microbial activity catalyzes oxygen transfer in membrane-aerated nitritating biofilm reactors. J. Memb. Sci. 2013, 446, 465–471.

527 528

(37)

Power, M. The predictive validation of ecological and environmental models. Ecol. Modell. 1993, 68 (1–2), 33–50.

529 530

(38)

Brockmann, D.; Rosenwinkel, K.-H.; Morgenroth, E. Practical identifiability of biokinetic parameters of a model describing two-step nitrification in biofilms. Biotechnol. Bioeng. 2008, 101 (3), 497–514.

531 532

(39)

Lackner, S.; Smets, B. F. Effect of the kinetics of ammonium and nitrite oxidation on nitritation success or failure for different biofilm reactor geometries. Biochem. Eng. J. 2012, 69, 123–129.

533 534 535

(40)

Vangsgaard, A. K.; Mauricio-Iglesias, M.; Valverde-Pérez, B.; Gernaey, K. V.; Sin, G. pH variation and influence in an autotrophic nitrogen removing biofilm system using an efficient numerical solution strategy. Water Sci. Technol. 2013, 67 (11), 2608.

536 537 538

(41)

Wyffels, S.; Van Hulle, S. W. H.; Boeckx, P.; Volcke, E. I. P.; Cleemput, O. Van; Vanrolleghem, P. A.; Verstraete, W. Modeling and simulation of oxygen-limited partial nitritation in a membraneassisted bioreactor (MBR). Biotechnol. Bioeng. 2004, 86 (5), 531–542.

539 540 541

(42)

Volcke, E. I. P.; Sanchez, O.; Steyer, J.-P.; Dabert, P.; Bernet, N. Microbial population dynamics in nitrifying reactors: Experimental evidence explained by a simple model including interspecies competition. Process Biochem. 2008, 43 (12), 1398–1406.

542 543

(43)

Chung, J.; Bae, W.; Lee, Y.-W.; Rittmann, B. E. Shortcut biological nitrogen removal in hybrid biofilm/suspended growth reactors. Process Biochem. 2007, 42 (3), 320–328.

544 545 546

(44)

Brockmann, D.; Morgenroth, E. Evaluating operating conditions for outcompeting nitrite oxidizers and maintaining partial nitrification in biofilm systems using biofilm modeling and Monte Carlo filtering. Water Res. 2010, 44 (6), 1995–2009.

547 548 549

(45)

Yoo, H.; Ahn, K.-H.; Lee, H.-J.; Lee, K.-H.; Kwak, Y.-J.; Song, K.-G. Nitrogen removal from synthetic wastewater by simultaneous nitrification and denitrification (SND) via nitrite in an intermittently-aerated reactor. Water Res. 1999, 33 (1), 145–154.

550 551 552

(46)

Gilbert, E. M.; Agrawal, S.; Brunner, F.; Schwartz, T.; Horn, H.; Lackner, S. Response of Different Nitrospira Species To Anoxic Periods Depends on Operational DO. Environ. Sci. Technol. 2014, 48 (5), 2934–2941.

553 554 555

(47)

Kornaros, M.; Dokianakis, S. N.; Lyberatos, G. Partial Nitrification/Denitrification Can Be Attributed to the Slow Response of Nitrite Oxidizing Bacteria to Periodic Anoxic Disturbances. Environ. Sci. Technol. 2010, 44 (19), 7245–7253.

556 557 558

(48)

Gilbert, E. M.; Agrawal, S.; Brunner, F.; Schwartz, T.; Horn, H.; Lackner, S. Response of Different Nitrospira Species To Anoxic Periods Depends on Operational DO. Environ. Sci. Technol. 2014, 48 (5), 2934–2941.

559 560 561

(49)

Li, J.; Elliott, D.; Nielsen, M.; Healy, M. G.; Zhan, X. Long-term partial nitrification in an intermittently aerated sequencing batch reactor (SBR) treating ammonium-rich wastewater under controlled oxygen-limited conditions. Biochem. Eng. J. 2011, 55 (3), 215–222.

562 563 564

(50)

Mota, C.; Head, M. A.; Ridenoure, J. A.; Cheng, J. J.; de los Reyes, F. L. Effects of Aeration Cycles on Nitrifying Bacterial Populations and Nitrogen Removal in Intermittently Aerated Reactors. Appl. Environ. Microbiol. 2005, 71 (12), 8565–8572.

565 566

(51)

Kornaros, M.; Marazioti, C.; Lyberatos, G. A pilot scale study of a sequencing batch reactor treating municipal wastewater operated via the UP-PND process. Water Sci. Technol. 2008, 58 (2), 435–438.

567 568

22 ACS Paragon Plus Environment

Page 23 of 27

569

570

Environmental Science & Technology

Table 1. Kinetic parameter values of AOB and NOB in the calibrated model.

1

Kinetic parameters

AOB

NOB

References

𝜇𝑚𝑎𝑥 : the maximum specific growth rate, 1/d

2.35 2.721

2.15 1.751

this study

𝐾𝑂2: half-saturation coefficient for O2, mg/L

0.6

1.2

Hiatt and Grady28

𝑌: autotrophic yield, mgCOD/mgN

0.18

0.06

Hiatt and Grady28

𝐴𝑂𝐵 𝑁𝑂𝐵 𝐾𝐹𝐴 , 𝐾𝐹𝑁𝐴 : half-saturation coefficient, mg/L

0.0075

0.0001

Hiatt and Grady28

𝐾𝐼,𝐹𝐴 : free ammonia inhibition coefficient, mg/L

1

0.2

Hiatt and Grady28

𝐾𝐼,𝐹𝑁𝐴 : free nitrous acid inhibition coefficient, mg/L

0.1

0.04

Hiatt and Grady28

𝑏𝑚𝑎𝑥 : decay coefficient, 1/d

0.17

0.073

Hao et al.34

pHopt (ω): optimal pH

8.4(3.2)

7.7(2.4)

Park et al.13

default growth rates in ASMN with temperature correction (33◦C)

571

23 ACS Paragon Plus Environment

Environmental Science & Technology

572

Page 24 of 27

Table 2. Predicted nitritation efficiencies (NE, %) in various intermittent aeration strategies

Influent2 Simulation Case

+

Effluent (Bulk) +

FA3

(mg-N/L)

Buffer capacity2

(mg-N/L)

continuous

75

2.1

39

6.96

0.27

0.01

A: 6+61

75

2.1

53.0 ± 5

7.23 ± 0.15

0.71

1.004

B: 1+1

75

2.1

52.5 ± 1

7.22 ± 0.02

0.69

0.73

C: 12+12

75

2.1

53.1 ± 10

7.25 ± 0.25

0.78

0.79

D: 8+4

75

2.1

47.8 ± 4

7.14 ± 0.15

0.52

0.41

E: 6+6

100

2.1

72.0 ± 7

7.25 ± 0.15

1.02

1.74

F: 6+6

50

2.1

35.0 ± 3

7.20 ± 0.15

0.45

0.21

G: 6+6

50

5

31.2 ± 5

7.41 ± 0.10

0.64

0.83

NH4

in

NH4

pH

(mg-N/L)

NEnormalized4

573

1

574

a 6-hour non-aeration phase. 2Buffer capacity in the influent was recorded as the molar ratio of bicarbonate

575

(HCO3-) to ammonium (NH4 _N). 3FA was calculated with the averaged NH4 concentrations and bulk pH

576

during a full aeration cycle (equation 6). 4For a clear comparison, NE was normalized to the Nitritation

577

efficiency in the default simulation case A (NE = 48.5%). MSNBM was run in continuous aeration (200 days)

578

to achieve a mature nitrifying biofilm, followed by various intermittent aeration strategies: (A-D) different

579

intermittent aeration but the same influent; (A,E-G) the same aeration intermittency but different influent

580

concentrations. NEs in the NOB suppression process in intermittent aeration were recorded (e.g. at day 215)

581

(Table S6). In simulations E-G, oxygen loadings proportionally varied with NH4 influent concentrations

582

(more simulations in Table S8).

Aeration strategy 6+6 meant a 12-hour intermittent aeration cycle consisting of a 6-hour aeration phase and

+

+

+

24 ACS Paragon Plus Environment

Page 25 of 27

Environmental Science & Technology

583 584

Figure 1. Experimental (discrete symbols) and predicted (line) concentrations in MABR at steady state (A)

585

microprofiles in the first aeration hour, (B) microprofiles in the last aeration hour, and (C) bulk profiles

586

during a 12-hour intermittent aeration cycle. For each micro profile, replicates (n>3) were made and the

587

average was shown.

588

25 ACS Paragon Plus Environment

Environmental Science & Technology

Page 26 of 27

589 590

Figure 2. A- Specific growth rates of AOB and NOB within the biofilm in a 6-hour aeration period at day 15

591

(AOB- black, NOB- red). B- Individual effect on AOB and NOB within the 100μm-aerated biofilm base in a

592

6-hour aeration period at day 15. (0- strong limitation/inhibition effect, 1- no limitation/inhibition effect)

593

26 ACS Paragon Plus Environment

Page 27 of 27

Environmental Science & Technology

594

595 596

TOC- Graphical abstract

27 ACS Paragon Plus Environment